[SensorNetArch] [Fwd: Re: FPS & Send/recv interface]

[Cheng Tien Ee]

Reply from Barbara...

-------- Original Message --------
Subject: 	Re: FPS & Send/recv interface
Date: 	Wed, 02 Mar 2005 12:08:28 -0800
From: 	hohltb at eecs.berkeley.edu
To: 	Cheng Tien Ee <ct-ee at EECS.Berkeley.EDU>



Hi Ee,
I am glad to hear you are working on this. FPS manages the SendQueue. Policy is separated from mechanism.

lib/Route and lib/MintRoute do not separate policy from mechanism so FPS cannot integrate with these routing protocols. The were not written to support the radio being off.

For example:
QueuedSend uses a naive send policy where once a message is put on the queue it is sent soon afterward.  (Alternatively, one can view it as MultiHopEngineM uses a naive policy where once it receives a message it calls QueuedSend.)

I think a good configuration for FPS would be to wire MultiHopEngineM to my new QueuedAllocSend and then have FPS implement its policy for QueuedAllocSend. 

I am working on the new configuration now and will check it in to tinyos-1.x/contrib/ucb/lib
CircleQueue
QueuedASend
SchedRoute

Best Regards,
Barbara

----- Original Message -----
From: Cheng Tien Ee <ct-ee at eecs.berkeley.edu>
Date: Monday, February 28, 2005 3:31 pm
Subject: FPS & Send/recv interface

> Hi Barbara,
> 
> I'm working on the interface between the network protocol and the 
> underlying MAC, wondering what sort of information should be 
> passed to 
> and fro. Is the current send and receive interface sufficient for 
> FPS to 
> run over CSMA? Is there anything in particular you wish the 
> interface 
> can provide?
> 
> Ee
>